Serverless computing may kill Google Cloud Platform
- 23 August, 2017 20:00
Google, which has had to claw its way back into cloud relevance in the shadows of Amazon Web Services and Microsoft Azure, suddenly finds itself playing catchup again, thanks to the rise of serverless computing. Although Google Cloud Platform still trails AWS and Azure by a considerable margin in general cloud revenue, its strengths in AI and container infrastructure (Kubernetes) have given it a credible seat at the cloud table.
Or would, if the world weren’t quickly moving toward a serverless future.
For years, AWS had the cloud infrastructure-as-a-service (IaaS) market all to itself. Microsoft was late to the party, but with its strong ties to CIOs and a savvy hybrid cloud story, Azure has quickly become a serious contender to Amazon’s cloud throne. Google, however, has had to settle for a distant third place:
As Gartner notes, Google Cloud Platform “is most attractive to cloud-native companies and those that want to ‘run like Google.’” Unfortunately for Google, most enterprises simply don’t fit either description, and although Google has tried to broaden its appeal to mainstream enterprises over the past year, those efforts have yet to generate AWS- or Azure-size bank balances. Indeed, Google delivers just a fraction—7 percent—of the cloud revenue that AWS does ($300 million compared to $4.1 billion last quarter, estimates Bernard Golden). Google Cloud Platform is also generating just 17 percent of Azure’s revenue.
Google faces the serverless destroyer
Serverless is still nascent, of course. Although AWS Lambda has been around since 2014, it’s still a relative minnow in the broader pond of cloud services. But such diminutive market presence belies the impact its having, particularly at the rate serverless computing has been maturing.
As Expedia vice president of cloud Subbu Allamaraju has declared, “Serverless patterns are pulling the rug from underneath container cluster managers faster than the latter [are] becoming industrial grade.”
Yes, you read that right. According to Allamaraju, serverless computing is outpacing industry darlings like Kubernetes. His own Expedia “did over 2.3 billion lambda calls per month” back in late 2016, a number that has climbed since then. Nor is Expedia alone in discovering the productivity gains to be found with serverless computing: Coca-Cola, Nordstrom, Reuters, and others have jumped in.
Yet it’s been AI and machine learning technologies like Kubernetes that Google has pinned its cloud hopes on. Focused on its Kubernetes-to-GCP and machine learning plays, Google has not built out the array of serverless services that its competitors have. As Mytton notes, “Once your core runtime requirements are met, the differences between the [serverless vendors’] services aren’t particularly important. … What does count is the availability of services to consume from within the cloud provider ecosystem.”
Even where Google has invested in serverless, it has done so half-heartedly, Mytton says:
Google Cloud Functions stagnated quite a bit between the alpha release in February 2016 and the beta release in March 2017. It still only supports a single runtime environment (Node.js) and seems to be suffering from App Engine syndrome —big announcements of alpha/beta features followed by silence/minimal progress until the next big announcement the following year.
Meanwhile, Google’s own Alan Ho contends that “from a programming model and a cost model, AWS Lambda is the future—despite some of the tooling limitations.” Those limitations, as Allamaraju speculates, are quickly disappearing, and in the process they may well put Google’s more traditional cloud to the sword.
We’re barely into the container era, and already serverless computing threatens to shorten its stay. Of course, Google (as well as AWS and Microsoft Azure) will collect plenty of cash for helping enterprises to modernize with containers, generally scaled out with Kubernetes. That will help Google make up some ground lost in the early cloud wars.
But unless Google can get its serverless act together, it may end up winning the container battle but losing the cloud war—which will increasingly be fought without containers at all.